Aluminum plasmonics for optical applications

(1)

HAL Id: tel-02965285

https://tel.archives-ouvertes.fr/tel-02965285

Submitted on 13 Oct 2020

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Aluminum plasmonics for optical applications

Dmitry Khlopin

To cite this version:

Dmitry Khlopin. Aluminum plasmonics for optical applications. Micro and nanotechnolo- gies/Microelectronics. Université de Technologie de Troyes, 2017. English. �NNT : 2017TROY0034�.

�tel-02965285�

(2)

(3)

THESE

pour l’obtention du grade de

D OCTEUR de l’U NIVERSITE DE T ECHNOLOGIE DE T ROYES

Spécialité : MATERIAUX, MECANIQUE, OPTIQUE ET NANOTECHNOLOGIE

présentée et soutenue par

Dmitry KHLOPIN

le 15 décembre 2017

Aluminum Plasmonics for Optical Applications

JURY

M. M. KOCIAK DIRECTEUR DE RECHERCHE CNRS Président Mme S. CAMELIO PROFESSEURE DES UNIVERSITES Rapporteur

M. W. DICKSON LECTURER Examinateur

M. D. GÉRARD MAITRE DE CONFERENCES Directeur de thèse M. J. MARTIN ENSEIGNANT CHERCHEUR UTT Directeur de thèse Mme A. PILLONNET PROFESSEURE DES UNIVERSITES Rapporteur

(4)

i

Acknowledgements

Looking back to the time I worked on my PhD thesis, I clearly understand that it could not be done without help of many people. Using this opportunity I would like to acknowledge everybody who gave their impact. First off all, I would like to express my deepest gratitude to my supervisors, Dr. Jérôme Martin and Dr. Davy Gérard. Thank you for your guiding during our work, for sharing your knowledge and experience within the different aspects, for the support and encouraging with my ideas. Your optimism, wisdom and patience were the source of my inspiration during this years.

I would like to thank my thesis committee members: Prof. Anne Pillonnet, Prof. Sophie Camelio, Dr. Wayne Dickson, and Dr. Mathieu Kociak for their time, advises and contribution.

My great thanks to our lab engineers Jérémie Beal, Sergei Kostcheev and Regis Deturche for technical support and advises in every moment. As well I would like to thank all my colleagues in LNIO for the precious time and great atmosphere. Especially my colleague Frédéric Laux for the fruitful discussions and help. Also our colleagues from King’s College in London take my gratitude for the great collaboration work.

I thank the team of École Doctorale, Isabelle Leclerq, Therese Kazarian and Pascale De- nis, for incredible patience and help.

I should mention my friends here in France and back in Russia. Thank you for your belief.

And of course, I would like to thanks my family for support, motivation and love. Your encouragement guided me to my place here.

(5)

(6)

iii

General introduction

Modern plasmonics is based on the intense and confined electromagnetic fields appearing near metallic nanostructures illuminated at frequencies near their localized surface plasmon resonances. The field of plasmonics is rapidly growing since the end of previous century.

Nowadays it proposes multiple applications for enhancing efficiency of photonic devices.

Among the different metals, aluminum sustains a broad range of surface plasmon resonances from deep UV to near IR due to its properties. Sharp plasmonic resonances can be achieved with aluminum nanostructures in near UV where popular noble metals cannot provide a comparable quality of resonance. Aluminum also has a native oxide protection layer and it is a very economically efficient material due to its low cost and recycling quality.

However, aluminum has high losses in the visible part of spectrum and its interband transitions are placed around 800 nm. This is why aluminum is usually less attractive for visible plasmonics in comparison with noble metals.

This thesis is a contribution to the wide field of aluminum plasmonics. The main goal consists in adapting the plasmonics techniques commonly used with noble metals in order to push them toward the blue and UV region using aluminum. This requires fabrication processes and a specific procedure for the fabrication of complex aluminum nanostructures, with applications in fields like light generation and extraction. A bibliographic summary of the state-of-the-art of the studied field is presented in Chapter 1. Theoretical aspects of general and aluminum plasmonics are described and several applications of aluminum structures for optics are discussed.

In the second Chapter, a strategy to increase the resonance quality based on diffractive coupling in periodic arrays is presented. Plasmonic resonances coupled with diffractive Rayleigh anomalies lead to hybrid modes with sharp resonances. This approach is studied theoretically and with FDTD simulations. Further samples were fabricated and investiga- tions based on achieved data were made.

Chapter 3 proposes an application of the studied arrays of nanoparticles. Aluminum nanostructures are coupled with a wide band gap semiconductor to enhance its emission.

Periodic arrays of Al nanoparticles were fabricated onto a ZnO epitaxial layer. Theoreti- cal concept of the sample contains several aspects of enhancement such as: Purcell effect, antenna effect and waveguiding. Their influence and the results are explained here.

Next Chapter derives from the problematics of Chapter 3: “How to increase the enhancement effect and get more efficient surface coverage without structuring semiconductor substrate?”. For this purpose we used a fractal geometry inspired by radiowave technology. The FDTD simulations were performed to design an effective geometry optimized

(9)

2 General introduction for certain wavelength. After the structures were fabricated with an adapted electron beam lithography process.

Finally, we propose a concept of chiral fractals. Using the complex geometry of fractals, it is possible to push optical chirality of plasmonic structures towards the UV part of the spectrum. This possibility was proved with the FDTD simulations in the second part of Chapter 3. Samples of chiral fractals were fabricated and the existence of circular dichroism in fractal structures was experimentally demonstrated.

(10)

3

Chapter 1 Aluminum plasmonics

The light lives in all places, in all things.

You can block it, even try to trap it, but the light will find its way.

The Speaker,Destiny 2

1.1 General word

A plasmon is a quanta of oscillations of the free electron gas (i.e. a plasma) in a metal. Its existence can explain several optical properties of metals. Their first applications appeared long before the name plasmon was even coined. Since ancient times, stained glasses were made by the inclusion of small metallic particles in glass. A well-known example is the Lycurgus cup [Freestone et al.,2007], a Roman cup dated from the IVth century and made from glass with metallic inclusions. Then, stained glasses were massively used in churches and palaces, like in the Cathedral of Saint-Pierre-et-Saint-Paul of Troyes. The first theory of the "plasma oscillation" was proposed by [D. Pines and D. Bohm,1952], with an experimental observation using electron energy loss spectroscopy by Powell and Swan in 1959 [Powell and Swan,1959], using aluminum thin films. From that point on plasmonics started to grow, however it is the rise of nanotechnology that will incredibly boost the field. It can be argue that the starting date of "modern" plasmonics is 1998 with the publication of T.W. [Ebbesen et al.,1998] in Nature about the extraordinary transmission of light through hole arrays. The main change brought by nanotechnology is the ability to manufacture nanostructures with controlled size and shape, such as the nanoholes in Ebbesen work. Nowadays plasmonics is featured in more than 1500 publications per year and thousands of applications. Currently it is one of the major driving forces of the fascinating field of nanooptics, which explores how electromagnetic fields can be confined and manipulated at the nanoscale.

1.1.1 Dielectric permittivity

Optical properties of materials, including metals, are characterized by a variable called dielectric permittivity –(ω). In this chapter we discuss the origin of this dependence.

(11)

4 Chapter 1. Aluminum plasmonics Maxwell’s equations are powerful tools of classic electrodynamics, which can be used to describe the electron gas appearing due of the huge electron density inside metals and negligible difference in energy levels in comparison withkT.

∇D~ =ρ_{f ree}

∇B~ = 0

∇ ×E~ =−∂ ~B

∂t

∇ ×H~ =~j+∂ ~D

∂t

(1.1)

whereE~ is the electric field,D~ - the electric induction,B~- the magnetic induction andH~ - the magnetic field.~jandρare current and charge densities respectively. Those values are connected through electric and magnetic polarization via the constitutive equations, presented for Fourier space for simplification:

D(k, ω) =~ 0(k, ω)E(k, ω) +~ P~(k, ω) B~(k, ω) = 1

µ0µ(k, ω)

H(k, ω) +~ M~(k, ω) (1.2) and

ρ=−∇P~

~j = ∂ ~P

∂t

(1.3) whereµ is the magnetic permeabity and is a dielectric permittivity. Both are generally tensors. In the case of a linear isotropic material with negligible space dispersion those equations take the following form:

D(k, ω) =~ ₀(ω)E(k, ω)~

B(k, ω) =~ µ₀µ(ω)H(k, ω)~ (1.4) Dielectric permittivity is a complex number=Re() +i·Im()and is directly connected to the complex refractive index:

˜ n=√

=n+iχ (1.5)

Parameterχhere is the extinction coefficient connected with absorption in a material. From Eq. 1.5 can be derived:

Im() = 2nχ (1.6)

This means that the imaginary part ofdescribes the absorption of materials.

(12)

1.1. General word 5 1.1.2 Band theory and origin of electron gas

The main reason behind the unique properties of metals lies in their electronic configuration. In a single metal atom, electrons have discrete energy levels known as atomic orbitals.

Usually, metals at standard conditions have less then 4 electrons on the outer orbit. For example, aluminum electronic configuration is 1s²2s²2p⁶3s²3p¹, which means 3 electrons in the outer orbit.

When two atoms are combined in a molecule, the atomic energy levels split into pairs of molecular energy levels. Accordingly, in a crystal structure the atomic levels split into a huge number of energy levels, creating continuous bands of energy states.

From quantum statistics, we have a Fermi-Dirac distribution that describes energy distribution of identical particles with half-integer spin, like electrons, in a system at thermo- dynamic equilibrium.

˜

n_i= 1

e⁽ⁱ^−µ)/k^B^T + 1 (1.7)

wherekis Boltzmann’s constant,T is the absolute temperature,iis the energy of the single- particle statei, andµis the total chemical potential. The probability that the many-particle system is in the stateR, is given by the normalized canonical distribution [Reif,1998]

P_R= e^−βE^R P

R⁰

E^−βE^R⁰q‘ (1.8)

whereβ= 1/k_BT From this the Fermi levelE_F can be derived as energy where probability of finding a particle becomes 50 percent:

E_F = ¯h²

2m(3π²N_e

V )^2/3 (1.9)

In metals with a given band structure, the Fermi levelEF lies inside at least one energy band. In semiconductors and dielectricsE_F is located in the band gap. This leads to the fact that inside of the band some energy levels a free, and it allows electrons to move along the material. For example in figure1.1presenting band structure of aluminum.

In summary, in metals electrons can freely move from one atom to another. Those electrons form the free electron gas with concentrationn₀inside of the metal.

Unique interaction between metal and light goes from oscillations of the electron gas.

1.1.3 Drude-Sommerfeld theory

P. Drude, 1900created a very efficient plasma theory to explain the optical properties of metals based on the interaction of light with electron gas. Later on, this theory was supple- mented by Lorentz and Sommerfeld, so the now in the literature Drude-Lorentz-Sommerfeld theory can be mentioned [Bartlett et al.,2004]. In the origin this model uses a definition of free electrons that combines non-uniformity of crystal lattice potential and electron-electron interaction, i.e. band theory in further study, into the effective mass of the electronm^∗. In this case Drude theory do not works well at the wavelengths of interband transitions in

(13)

6 Chapter 1. Aluminum plasmonics

FIGURE1.1: Aluminum band strucuture. Extracted from Levinson, Greuter, and Plummer,1983.

metals, because within this region band influence cannot be neglected. Within the Drude model, the equation governing the movement of an electron in the presence of an external electric fieldEtakes the form:

m^∗x¨+m^∗γx˙ =−eE (1.10)

whereγ = 1/τ is the frequency of collisions of electrons andτ is a free path time, xis the position, (¨x is the acceleration,x˙ is the speed, eis the electron charge and E is the electric field.

FIGURE1.2: Drude model electrons (shown here in blue) constantly bounce between heavier, stationary crystal ions (shown in red). Extracted from Wiki-

media.

(14)

1.1. General word 7 In case of a monochromatic, time-harmonic fieldE(t) =E0e^−iωt)solution becomes:

X(t) = eE

m^∗(ω²+iωγ) (1.11)

Shift of the electron gas leads to the polarization:

P =−n₀eX =− n0e²E

m^∗(ω²+iωγ) (1.12)

From this equation in combination= 1 +χ, which comes from eq. 1.5 we can derive the dielectric permittivity:

(ω) = 1− ω_p²

ω²+iωγ (1.13)

whereω_p =p

n₀e²/m^∗ - the plasma frequency of free electron gas. At high frequencies, where ωτ >> 1imaginary part of can be neglected and ≈ 1−ω_p²/ω². But in the area of interband transitions real and imaginary parts become roughly of the same order. As a compensation, in this case, a Drude-Lorentz model can be used as follows:

(ω) =_Drude(ω) +_Interband(ω) = (1− ω_p²

ω²+iωγ) + ( G₀ω²₀

ω²₀−ω²−iΓω) (1.14) where ω0 is the central frequency, G0 is a gain, andΓ the damping factor from Lorenzian model of the interband transitions.

This approximation also works well for macro particles, in classic electrodynamics. But if we decrease the size of the particle to the nanometric scale, free mean path of electrons become comparable with the size of the particle. This also will increase imaginary part of the dielectric permittivity as soon as it is dependent fromγ. With a decrease of any dimension below ≈10 nm, quantum effects and separated energy levels start to affect the properties and the model should be much improved. As example of values following1.1is provided:

Au Ag Al

no(m⁻³) 5.9∗10²⁸ 5.76∗10²³ 2.1∗10²⁹

ωp(eV) 9.1 9.1 15

τ(f s) 27.3 36.8 11.8

TABLE1.1: Drude parameters for gold, silver and aluminum. Extracted from Klimov,2014, Ordal et al.,1983and Gall,2016.

Now assume situation with infinite metal volume. Dispersion law for transversal waves (div ~E = 0) will take form:

ω²(ω) =c²k² (1.15)

And in case of Drude dispersion:

(ω) = 1−ω_p²

ω² (1.16)

(15)

8 Chapter 1. Aluminum plasmonics From those equations, the existence conditions of a wave, the so-called volume plasmon, can be derived:

ω= q

ω²_p+c²k² (1.17)

The dispersion curve of those conditions represented in figure1.3.

FIGURE1.3: Dispersion curve for volume plasmon. Extracted from White, 2007.

1.1.4 Surface plasmons

Previously we discussed volume plasmons. This type of plasmon are not localized waves, because their dispersion curve is located higher than the light line, see Fig. 1.3. Much more interesting for nanotechnology, surface plasmons can appear at the interface between a metal and a dielectric. The first mention of such oscillations appeared in works of [Zen- neck,1907] and [Sommerfeld,1909], connected with problems in telegraph. To describe the surface plasmon phenomena, we consider a simple metal-dielectric interface, as shown in figure1.4

In this configuration, solutions of the Maxwell’s equations can be written for the metal:

E~m=E~0{1,0, ksp

k_zm}exp{i(k_spx−kzmz−ωt}

H~m =−E~0

m

kzm

(ω

c){0,1,0}exp{i(k_spx−kzmz−ωt}

(1.18)

and for the dielectric side:

(16)

1.1. General word 9

ε

m x

z

ε

d

Dielectric Metal

FIGURE1.4: Metal-dielectric interface geometry.

E~_d=E~₀{1,0,−k_sp

k_zd}exp{i(k_spx+k_zmz−ωt}

H~_d=E~₀ _d k_zd(ω

c){0,1,0}exp{i(k_spx+k_zmz−ωt}

(1.19)

where

k_zm= r

_m(ω)(ω

c)²−k_sp² k_zd=

r

_d(ω)(ω

c)²−k²_sp

(1.20)

are the wavevectors orthogonal to the interface in metal and dielectric side correspondingly, and k²_sp - surface plasmon wavevector. From the continuity of tangential components of electric and magnetic fields, we can write the dispersion equation:

_m kzm

+ _d kzd

= 0 (1.21)

This equation shows the possibility to have a wave without a source, i.e. it is an electromagnetic mode. In combination with the previous equation, we get:

ksp= ω c

s

m(ω)d(ω)

m(ω) +_d(ω) (1.22)

And through this it possible to write the components of the wavevector that are orthogonal to the interface:

k_z,j² = (ω c)² ²_j

m+_d, with j =m, d (1.23)

(17)

10 Chapter 1. Aluminum plasmonics Surface plasmons are localized waves along the interface. It means that along the interface k_spwavevector should be real, andk_z,jshould be imaginary. This implies:

m(ω)d(ω)<0

m(ω) +d(ω)<0 (1.24)

Metals like gold, silver or aluminum have a large negative real part of permettivity and relatively small imaginary part. In this case surface plasmons can exist on the interface of those materials.

ω > ωp

ω < ω_p

√_d+ 1

(1.25) Figure1.5shows the dispersion curve from equation 1.19 if we take Drude’s dispersion for the metal_m(ω) = 1−ω_p²/ω². Two branches corresponding to volume and to surface plasmons as predicted by equation 1.22. Note that due to their nature, surface plasmons propagating along a metal-dielectric interface are sometimes called "propagating surface plasmon" or "delocalized plasmons". Another kind of plasmonic excitation, known as localized surface plasmons, also exists and is described in the next section.

FIGURE1.5: Dispersion curve for volume and surface plasmon. Extracted from White,2007.

1.1.5 Localized surface plasmons

Oscillation of electrons in metallic nanoparticles whose size is smaller than the wavelength of the exciting light are called localized surface plasmons. The main difference with volume and surface plasmons lies in the fact that localized surface plasmons can exist only at

(18)

1.1. General word 11 fixed resonance frequencies, depending on the particle’s shape, size and material. Those frequencies are called plasmonic resonances.

There are two main approaches for the theoretical calculation of plasmonic resonances and the associated distribution of electromagnetic fields:

1. The first method is to solve Maxwell’s equations, or their integral form, for different frequencies and to find the frequency-dependence of parameters of interest, such as the scattering cross-section. The scattering cross-section is maximum at the plasmonic resonance and plotting the electromagnetic fields at this frequency yields the spatial distribution of the plasmon mode. The main drawback of this method is that it cannot be easily applied to the different materials, even with the same particle form. More- over, the scattering cross-section is affected by material losses and other things, so it could decrease precision of calculations.

2. The second method, called the effective medium theory [Elser et al., 2006] is based on a important value, already introduced - dielectric permittivity, which depends on frequency. This method is more universal and allows a deeper understanding of the localized plasmons. In this method, a lot of attention is payed to the form of the particle and its influence on the optical properties.

Effective medium theory is well explained in the book of [V. Klimov,2014]. Using non- standard methods, solutions of Maxwell’s equations can be written in the form:

E =E₀ =X

n

e_n ((ω)−1) (_n−(ω))

R

V+

enE0dV R

V+

e²_ndV (1.26)

whereE₀ is the electric field without the particle,nis the mode index,_nande_nare eigen- value and eigenfunction, respectively, V+ is the volume inside the particle. There is a resonance factor _n−(ω). ω_n - resonance frequencies, leads to _n−(ω) = 0. Width of the resonance is dependent on imaginary parts ofnof(ω), but usuallyIm(n)<< Im((ω)).

An analytic solution can be derived for particles exhibiting a high degree of symmetry, such as spheres [Mie, 1908] or spheroids [Yamashita, Wakoh, and Asano, 1975]. Analytic solutions can be found for other forms but they become unnecessarily complicated and nu- merical solutions are generally used instead.

For particles with sizesd << λthe quasi-static approximation can be used. For example, let us consider the simple case of a homogeneous sphere of radius a, made from a Drude metal, inside a dielectric medium with permittivity m. A static electric field E = E0zˆis applied to the system. A sketch of the system is shown in figure1.6.

Potentials insideΦinand outsideΦoutcan be written as follows, derivation can be found in [Maier,2007] or [Jackson and Fox,1999]:

Φ_in=− 3_m

+ 2_mE₀rcosθ Φ_out=−E₀rcosθ+ −_m

+ 2_mE₀a³cosθ r²

(1.27)

(19)

FIGURE 1.6: Sketch of a homogeneous sphere placed into an electrostatic field. Extracted from Maier,2007.

To simplify those equations an induced dipole momentpis introduced:

p= 4π₀_ma³ −_m

+ 2_mE₀ (1.28)

And from there a polarizabilityα, a key feature of those calculations, can be defined as:

α = 4πa³ −m

+ 2_m (1.29)

From polarizabilityαwe can see that resonance takes place at(ω) =−2_m. Also from further calculations [Lord, Zhan, and Pawliszyn,2012] a cross section of scatteringCscaand absorptionC_absunder plane wave illumination can be found as follows:

C_sca = k⁴ 6π|α|² C_abs=kIm(α)

(1.30) wherek= 2π/λ.

1.2 Aluminum plasmonics

During the last decades, plasmonics has mainly involved the study of gold and silver nanostructures. This is primarily due to their suitable properties in the visible part of the spectrum, i.e. electric permeability has a negative real value and imaginary is relatively small.

But with development of the plasmonics field, the need for new materials became a cru- cial question due to the high price of noble metals and the limited covered part of spectra.

For a long time aluminum was underestimated for plasmonics purposes, both due to its relatively high losses in the visible region and to its rapid oxidation when exposed to air.

However, as fabrication technologies and know-how increase, aluminum now appears as a great alternative material for plasmonics. Furthermore, aluminum provides an economic advantage because it is cheap and widely available, and compatible with complementary metal–oxide–semiconductor (CMOS) technology.

(20)

1.2. Aluminum plasmonics 13 1.2.1 Basic optical properties of aluminum

As discussed previously, aluminum has a high concentration of free electronsn = 6·10²³ m⁻³. The plasma frequencyω_p directly depends on this concentration. For aluminum, the plasma frequency is ωp = 15eV. That is≈ 60%higher than the same values for gold and silver (see table1.1). So by the formula 1.22 we can assume that aluminum should provide a wider spectral range for surface plasmons than gold or silver. But since pioneering studies back in the middle of the 20th century [Ehrenreich, Philipp, and Segall, 1963], aluminum has the reputation of being a lossy and unstable material, not suitable for plasmonics. As a consequence, it has not been studied widely as a plasmonic material until recently.

The main problems associated with aluminum can be summarized in a two very important issues:

1. Aluminum exhibits an interband transition around a wavelength of 800 nm, right in the middle of a region of interest for the evolving field of plasmonics. Interband transitions creates high optical losses and also the classical Drude model approximation is not working properly in this region.

2. Aluminum has a high imaginary part offor wavelengths higher than roughly 500 nm and this means high losses in this region. Real and imaginary parts of the dielectric permittivity can be observed in figure1.7.

FIGURE 1.7: Real and imaginary parts of the electric permittivity of aluminum. Extracted from Gérard and Gray,2015. Data was taken from work of

Palik,1997and Raki´c,1995.

However, those disadvantages can be turned into advantages. Silver and gold have a lower frequencyωpand both of them have interband transitions in blue and near-ultraviolet part of the electromagnetic spectrum. In this region, aluminum overcomes all noble metals in optical properties. Group of [West et al., 2010] performed simulations with finite difference time domain method (FDTD) of the quality factor for plasmonic resonances of spherical particles made out of different metals. Results of their work shown in figure1.8. As can be seen, aluminum shows much higher values in the UV region than the others.

(21)

FIGURE 1.8: Quality factors for localized surface plasmon resonances (Q_{LSP R})is shown in panel (a), and surface plasmon polaritons quality fac-

tor(Q_{SP P})is shown in panel (b). Extracted from West et al.,2010.

1.2.2 Oxidation of aluminum

Even outside the field of optics, aluminum is well known for its lightweight and resistance to erosion. The second property comes from the fact that top layer of aluminum oxidizes incredibly fast, forming an alumina (Al2O₃) layer. Alumina is a very robust material with a high melting point. The solid alumina layer also affects any fabrication process – this will be discussed later. Fabricated aluminum nanostructures oxidize right after fabrication by just being in air. However, after forming a 3 nm native layer of alumina, the oxidation process stabilizes and particles can keep the same oxide shell thickness for a long time: from a week to months, depending on the environmental conditions (humidity and temperature)[Langhammer et al.,2008a].

Oxidation is of course detrimental for plasmonic properties. It creates a gap between the surface of the particle and the metal. For some applications this could be critical, for example for Surface Enhanced Raman Spectroscopy (SERS). In figure1.9presented simu- lated dependence of extinction spectra versus the thickness of alumina shell for constant size of the particle [Zhang et al.,2017]. As found in this interesting work, there is an obvious decrease of half-width of the resonance peak, i.e. quality factor (Fig1.9a), but also both a red and a blue shift are observed (1.9b). The red shift comes from a high refractive index of alumina surrounding the metal core, while the blue shift appears due to the decreasing the size of the core, eventually overcoming the red shift.

The main topic of the work described by F. Zhang is about increasing quality of plasmonic resonances through Rapid Thermal Annealing (RTA). One of the biggest issues with aluminum particles fabrication is their poly-crystalline grain structure. Several methods have been reported [Martin and Plain, 2015] to improve fabrication using either low or high evaporation speeds of the metal, but mono-crystalline aluminum is unachievable with simple methods such as thermal evaporation. Using a RTA post-treatment on aluminum nanoparticles, F. Zhang achieved a bigger grain size with moderate oxide layer growth.

This leads to increased quality factors of the localized surface plasmon resonance (LSPR), as

(22)

1.2. Aluminum plasmonics 15

FIGURE1.9: Extinction spectra calculated with FDTD for a square array of Al nanoparticles with diameter d = 100 nm and thickness h = 40 nm for different ratios between the volume of Al core and the volume of oxide layer (from oxide thickness L = 0 nm, no oxidation, to L = 30 nm). (b) Corresponding extinction spectra for several oxide thicknesses. Extracted from Zhang et al.,

2017

observed in figure1.10.

As a conclusion on the oxidation topic, we could say that oxide layer of alumina is not a real issue. Oxidation creates a robust thin native protection layer, which can also can be used as a resonance tuning instrument or nanosized melting form.

1.2.3 Aluminum nanoplasmonics

The basic approach to study the optical properties of aluminum nanotructures is to perform extinction spectroscopy on nanostructures with simple geometries. For example, there is a nice study of aluminum nanodisks and nanorods made by M. [Knight et al., 2014]. Figure 1.11reprinted from their article presents the evolution of the plasmonic resonance with increasing sizes. As it can be seen, with the shift of the resonance from UV to visible part of the spectrum, the shape of the resonance changes a lot – from a well-defined peak to a broad lowered peak. This also means that the quality factor is decreasing.

Those results are mainly explained by the increase of the electric permittivity of aluminum in the visible region. The grain structure also affects the resonance, due to the electron scattering at grain boundaries. With increasing particle size, this effect is more pronounced.

Although electron beam lithography is probably the best fabrication method for sys- tematic study, it is still difficult to fabricate aluminum particles with diameter lower than 40-50 nm. With those limitations and depositing the nanoparticles onto a quartz substrate, the lower boundary for the achievable resonance wavelength is around 280-300 nm. To go beyond this limit, completely different chemical methods should be used. For instance, in the work by [Maidecchi et al., 2013] particles were fabricated using self-assembly on a silane layer . Another example a quite recent method of chemical fabrication of aluminum

(23)

FIGURE1.10: Column (left), for each box: extinction spectra of Al nanoparticles with fixed diameter after successive RTA treatments with increasing temperature. Column (right): evolution of the peak wavelength and line width

with RTA treatment. Extracted from Zhang et al.,2017

nanoparticles with diameter down to 5 nm immersed in ethylene glycol described in patent [Plain et al.,2014]. It is important to mention that at such small sizes quantum effects might become important as discussed previously.

The other possibility to decrease the operating wavelength is to consider higher-order plasmonic modes. For example, quadrupolar plasmon resonances, whose resonance frequencies are higher than that of dipole resonances, can be used. Due to the properties of material, quadrupole and higher modes are more easily excited in aluminum when com- pared to noble metals, but still the associated absorption and scattering cross-sections are generally quite low. Under some circumstances it is possible to enhance the excitation of a quadrupolar resonance. The easiest way to do that is to use oblique illumination. In the work of [Martin et al.,2014] multipolar resonances was observed in aluminum nanoparticles with electron energy loss spectroscopy.

In most of the aforementioned works, scientists were trying to push the resonance wavelength toward the ultraviolet region. In this area aluminum has little concurrency from the other metals and shows remarkable optical properties. Moreover, and most importantly, UV-plasmonics opens the way to a huge amount of potential applications that we are going to discuss in the following.

(24)

1.3. Applications of aluminum plasmonics 17

FIGURE1.11: Tuning of the LSP resonance of Al nanodisks fabricated using e- beam lithography. (a) Experimental scattering spectra of individual nanodisks with varying diameters. (D = 70, 80, 100, 120, 130, 150, 180 nm) (b) SEM images of the corresponding nanodisk structures. Scale bar is 100 nm. (c) Finite difference time domain (FDTD) simulations of the scattering spectra, assuming a 3 nm surface oxide and a SiO2 substrate. Extracted from Knight

et al.,2014

1.3 Applications of aluminum plasmonics

The optical properties of aluminum described in the previous section show a great potential for application in different areas. The most promising way is near-UV and blue plasmonics.

Sensors, light sources, amplifiers in this region can be enormously useful for application and implementation with biology. Many important biomaterials are somehow sensitive to the UV light, and with aluminum this effect can be greatly increased. Moreover aluminum has low toxicity and, due to the protection of the native oxide layer, resistant to the environment.

1.3.1 Light generation

It is known for a while that surface plasmon can be coupled to luminescent materials, whether fluorophores or semiconductors, in order to enhance their emission. This field is known as "metal-enhanced fluorescence". Fluorescent molecules are more important

(25)

18 Chapter 1. Aluminum plasmonics for detection and investigation in biological and medical research [Liu et al., 2015], while semiconductor-metal structures and their derivatives (quantum dots, wires and wells) are mainly oriented toward the fabrication of light sources and detectors [Okamoto et al.,2004].

As a combination of both, there is a topic with a fast growing popularity: theranostics [Yu, Park, and Jon,2012]. The word "Theranostics" is a combination of "Therapy" and "Diagnos- tics", so as follows from the name this research topic is concentrated on creating heteroge- neous particles exhibiting multitask properties.

When a metal is combined with any light-emitting material, there are three main effects that may modify light generation:

1. At the plasmonic resonance, the local electric field intensity is increased in the vicinity of the particle, leading to an increased excitation rate;

2. The so-called Purcell effect makes the particle working as an optical resonator, speed- ing up relaxation rates in the active material. This yields higher emission rate;

3. Designing the shape of the particle allows one to increase the coupling between the far- field and the near-field, channeling emitted light more efficiently towards the detector or, conversely, concentrating impinging light on the active material. This is the antenna effect.

Aluminum as material for ultraviolet plasmonics shifts all these applications toward the UV region, which is interesting as many molecules can be excited in UV. Usually this emission has a low intensity. As example of such application, the work of J. Eid et al., 2009can be mentioned. DNA sequencing is a very hot topic, and numerous solutions to perform the sequencing have been proposed. One technique for detection of parts of cut DNA, i.e. nucleotides, is to attach fluorescent markers to them. In [Eid et al.,2009], nanosized apertures milled in a thin aluminum film (called "zero-mode waveguides") were used to enhance the signal emitted by such markers. A visual description of the method is shown in figure1.12.

Aluminum can be also used for enhancing of luminescence from rare earth materials, which absorb light in UV a emit in the UV-visible. Let us mention for example the work of Abdellaoui et al.,2015, where the coupling between Eu³⁺rare earth emitters and aluminum nanostructures has been investigated.

Concerning applications with semiconductors, an attractive direction is the coupling with wide band-gap semiconductors, such as zinc oxide (ZnO) with band gapEg = 3.37 eV, gallium nitride (GaN) withE_g = 3.2eV and boron nitride withE_g= 5.9eV. An interesting fact is that highly n- or p-doped ZnO and AlZnO semiconductors can have a plasmonic behavior at radio frequencies Boltasseva and Atwater,2011.

The simplest configuration of an Al-semiconductor structure consists of an aluminum film with a layer of semiconductor on its top (or reversed). In this configuration, the semiconductor can be a quantum well [Okamoto et al.,2004] [Gao et al.,2012] or the bulk material [Chou et al.,2015]. In this case the Al layer works as a mirror, amplifying the excitation

(26)

1.3. Applications of aluminum plasmonics 19

FIGURE1.12: Experimental geometry. A single molecule of DNA polymerase is immobilized at the bottom of a Zero-Mode aluminum waveguide (ZMW), which is illuminated from below by laser light. The ZMW nanostructure provides excitation confinement in the zeptoliter (10₋₂₁ liter) regime, enabling detection of individual phospholinked nucleotide substrates against the bulk solution background as they are incorporated into the DNA strand by the

polymerase. Extracted from Eid et al.,2009.

light, and also enhances the emission. The reported values of the experimental enhancement factor in those articles is more than 2 times. The next level is structuring the metal surface [Norek, Łuka, and Włodarski,2016] or changing it to the particle patterns. This will be discussed more precisely in chapter 3.

Using smaller structures to increase light confinement, we can obtain even more localized effects with higher impact. For example, the group of Q. [Zhang et al.,2014] showed a GaN nanolaser and [Chou et al.,2015] demonstrated a ZnO nanolaser. Both projects were based on a configuration where a semiconductor nanowire lies onto either an aluminum [Zhang et al.,2014] or a silver surface [Chou et al.,2015]. This creates a huge local electric field and Purcell effect from the interface metal-semiconductor as shown in figure1.13. In both works lasing from nanowires was achieved due to plasmonic enhancement.

FIGURE1.13: Robust ultraviolet ZnO nanolaser that can operate at room temperature by using silver to dramatically shrink the mode volume. Extracted

from Chou et al.,2015

(27)

20 Chapter 1. Aluminum plasmonics 1.3.2 Raman spectroscopy

Enhanced Raman spectroscopy is a powerful tool for material study through phonon sig- nature. Raman scattering, the core feature of this method, is an inelastic (Stokes) scattering where the energy lost by the photon is transferred to phonons. Raman scattering happens along with elastic (Rayleight) scattering, but with an extremely low intensity in comparison. In order to enhance the signal, surface plasmons in metals are widely used in different configurations. The two main directions of development in this field are:

1. Surface Enhanced Raman Spectroscopy (SERS) relies on rough metallic films or nanostructures to intensify the Raman signal. Under the incident light excitation, the local electric field around the particles is increased, increasing in turn the Raman scattering (as the Raman scattering cross-section is roughly proportional to the fourth power of the electric field). This technique allows the probing of large surfaces as well as single molecule located in a hot spot between particles [Vlckova et al.,2007]. A sketch is shown in Fig. 1.14a;

2. In Tip Enhanced Raman Spectroscopy (TERS), a metallic (or metal-covered) sharp tip is used. It is needed to create a local increase of the Raman signal at its apex. Combined with a conductive surface, it allows one to perform scanning tunnelling microscopy (STM) at the same time. The sharpness of the tip also provides the possibility to mea- sure the local Raman signal, hence creating a map of the Raman signal with nanoscale resolution [Sheremet et al.,2016]. Schematics can be seen in Fig.1.14b.

FIGURE 1.14: (top)Conceptual illustration of SERS. Extracted from Surface Enhanced Raman Specrtoscopy. (bottom)Conceptual illustration of TERS. Ex-

tracted from Pettinger,2010.

The native oxide layer around aluminum nanostructures decreases the effectiveness of SERS but still, in the blue-UV region there is no better metal to use. In [Taguchi et al., 2009], the authors presented results for both methods: SERS using an 10-nm-thick aluminum film on a quartz substrate, and TERS using a silicone tip covered with 25 nm of aluminum.

Aluminum plasmonics for optical applications

HAL Id: tel-02965285

https://tel.archives-ouvertes.fr/tel-02965285

Aluminum plasmonics for optical applications

Dmitry Khlopin

To cite this version:

THESE

D OCTEUR de l’U NIVERSITE DE T ECHNOLOGIE DE T ROYES

Dmitry KHLOPIN

Aluminum Plasmonics for Optical Applications

JURY

Acknowledgements

Contents

General introduction

Chapter 1

Aluminum plasmonics

1.1 General word

ε

ε

1.2 Aluminum plasmonics

1.3 Applications of aluminum plasmonics